Letting LLMs write a tiny DSL instead of JSON tool calls: is it worth it?
bye_ella · reddit · 2026-11-14
A Reddit discussion explores an agent design alternative: instead of step-by-step JSON tool calls, give the model a small DSL to write a full task program run by a validated interpreter. Pros: far fewer output tokens, no per-step round trips, grammar-constrained decoding even for small/quantized models, errors caught before execution. Open questions: models may reason worse in an unfamiliar DSL, and whole-program generation can't adapt to intermediate results. The poster asks for real production experience and whether to mix approaches.
More from coding & agent
- NetBird ships an MCP server so Claude Code and Cursor can manage your network — philipvollet · 2026-11-14
- MemCredit: Semantic Relevance Fails at Agent Memory Selection, Paper Finds — omarsar0 · 2026-11-14
- Train system design requirement-discovery with AI using vague-problem drills — anirudhology · 2026-11-14
- Dev with zero KiCad experience gets Codex to do PCB design in half a day — mckbrando · 2026-11-14
- One prompt, one 97KB HTML file: a procedural swimmable underwater world built by Claude — Any-Reputation8118 · 2026-11-14
- AG-UI, the agent-user interaction protocol for frontend apps, hits 17k GitHub stars — ag-ui-protocol · 2026-11-14